Activities to encourage two-word phrases in late talkers

Last updated 2026-07-11

Most kids start putting words together, "more milk," "daddy go," somewhere between 18 and 24 months, according to the American Speech-Language-Hearing Association ^[1]. By age 2, expressive vocabulary typically runs 200 to 300 words, and two-word combinations are already a regular part of how a toddler talks ^[2].

The range is wider than most parents expect, though. Some children get there at 18 months; others don't manage it until 26 or 27 months and still catch up fine on their own, no help needed. Pediatric speech-language pathologists tend to start paying closer attention when a 24-month-old has fewer than 50 words and isn't combining any of them. That's usually the cue for an evaluation ^[1].

Late talkers aren't one group with one cause. Limited language exposure, a motor-planning difference like childhood apraxia of speech, an underlying profile such as autism, or something nobody's pinned down yet: any of these can sit behind a slow start. The good news is the activities that help tend to work no matter the cause. Still, if you're worried, it's worth doing both at once: practicing at home and getting a formal speech therapy evaluation, rather than picking one over the other. Research on late talkers also keeps landing on the same point: waiting does less good than stepping in early with targeted input. A systematic review by Roberts and Kaiser in the American Journal of Speech-Language Pathology found that parent-implemented language interventions led to statistically significant gains in expressive vocabulary and word combinations, compared with children who got no treatment ^[3].

Why kids get stuck at single words

Moving from single words to two-word phrases takes more than having the vocabulary for it. A child has to hold two pieces of information in mind, put them in the right order, and get them out together in real time. That's a bigger lift, mentally and physically, than it looks like from the outside.

For some kids the real bottleneck is vocabulary size: combinations tend to show up once a child has around 50 words ^[2]. Below that, the priority is still building single words, not pushing toward phrases before the foundation's there.

Other kids already have plenty of words but haven't figured out that combining them works better. Saying "more" gets the job done well enough for them. Nobody's shown them yet that "more juice" works even better.

Motor planning can be behind it too. Kids with apraxia of speech often know exactly what they want to say but struggle to sequence the sounds and syllables a longer phrase needs ^[4]. The same activities that help other kids still apply here, the pacing and cues just need to look a little different.

And then there are children who understand far more than they say. When a child's understanding runs well ahead of their speech, that's actually encouraging: the concepts are already in place. What's missing is a bridge to get them out.

What actually works, according to research

Three approaches keep showing up in the research on parent-led language help, and they all operate the same way: the adult talks, models, and expands, without ever putting the child on the spot.

The first is modeling without demand. You say the phrase you want your child to pick up in a natural moment, you don't ask them to repeat it, and you move on. This low-pressure approach sits at the center of the Hanen Centre's More Than Words program, which has a solid track record for children with autism and late talkers alike ^[5].

The second is milieu teaching, which sounds more clinical than it is. It just means weaving language into everyday moments and expanding slightly on whatever your child attempts. If a child reaches for a cracker and says "cracker," you hand it over and say "want cracker," naturally, no quiz attached.

The third is aided language stimulation (also called aided language input), where you point to symbols or pictures as you talk. This matters most for kids who aren't yet reliable verbal communicators. Exposure to AAC devices and symbol-based communication tends to speed up spoken language rather than get in its way. Research on aided language input published in Augmentative and Alternative Communication found gains in both symbol use and spoken language among children with complex communication needs ^[6].

Here's how the three line up against the evidence:

All three share one habit: the adult does most of the talking, most of the time, and the child never gets corrected for getting it wrong.

What are the best activities to encourage two-word phrases at home?

None of these require a worksheet. They fold into whatever you're already doing during the day.

Sabotage play is exactly what it sounds like: you set up a spot where the child has to ask for something to get it. Blow a bubble, then wait. Don't hand over the jar. They'll say "more" or "bubble." You say "more bubbles" and blow one. No quiz, no pressure, you've just modeled the combination and rewarded the request in the same breath. SLPs call this a communication temptation, and it's one of the highest-yield moves in early language work ^[5].

Predictable book routines work well too. Pick something with repeating lines, like "Brown Bear, Brown Bear" or "We're Going on a Bear Hunt," anything with a pattern. Read it enough times that the child can guess what's next, then pause right before the phrase and leave the space open. The child fills the gap, first with single words, then, over weeks, with the whole phrase. You're not drilling, you're making the pattern impossible to resist.

Try offering two-word choices instead of open-ended questions. "What do you want?" asks a lot of a small brain. Hold up two things instead and model both with full phrases: "Blue cup? Red cup?" The child points or reaches, you hand it over and say the phrase again. The choice structure drops the load and shows the exact form you want them to use.

Expanding everything and demanding nothing is less an activity and more a habit worth building. Whatever the child says, add one word. "Dog" becomes "big dog" or "dog run" or "silly dog." No correction, just a preview of what comes next. Research keeps showing that expansions one level above a child's current mean length of utterance are the most useful input ^[3].

Sensory bins give you a natural excuse to narrate. Fill a bin with rice, water, beans, or sand, play alongside your child, and describe what's happening in two-word phrases: "pour out," "wet sand," "dig more." You're modeling, not instructing. The child can copy you or not, but either way they're soaking in phrase-level language at a moment when they're relaxed and engaged.

Daily routines carry the same weight without any extra setup. Snack, bath, getting dressed: these happen every day and come with predictable words. "All done." "Wash hands." "Wet hair." "More please." Lock one phrase to each step of a routine and use the same one every time. Repetition is what does the work here; novelty is overrated for early language learning.

Cause-and-effect toys close the loop nicely. Pop-up toys, wind-up cars, a jack-in-the-box: the child sets it off, then looks at you because they want it again. That's your moment. Say "push button" or "do more" and help them do it. The motivation is already built in.

Running one of these sessions is looser than it sounds. Get down to your child's level, whether that's the floor or a low chair, so you're face to face. For the first couple of minutes, just follow along: no redirecting, no new toys, no questions. Narrate what they're doing. "Roll car." "Big tower." "Car go."

When your child communicates something, whether it's a word, a sound, a point, or a reach, respond to what they mean first, then add one word on top. Don't ask them to repeat it back. Just say it naturally and keep playing.

Aim for three comments or expansions for every question you ask. Questions put kids on the spot; comments just hand them language without any pressure to perform.

If your child isn't saying anything back, keep going anyway. Children who aren't talking yet still take in this kind of input, and the payoff sometimes doesn't show up for weeks ^[3]. And wrap up before interest fades: boredom shuts down language learning as fast as frustration does, so ten minutes of real engagement beats forty minutes with a kid who's checked out.

Will sign language or AAC get in the way of talking?

No, and this is probably the worry parents raise more than any other. The research is pretty settled: signs and AAC don't delay spoken language, and if anything they support it ^[6].

The worry makes sense on its face. If a child can get what they want by pointing to a symbol, why bother talking? But that's not how it plays out. A child with a reliable way to communicate is less frustrated and more engaged, and that tends to make them more likely to add spoken words alongside whatever other method they're using, not abandon speech for it.

ASHA's guidance on augmentative and alternative communication says this directly: AAC shouldn't be held back until other approaches have failed, and introducing it early doesn't get in the way of speech later ^[7].

This matters most with two-word combinations. AAC lets a child practice putting phrases together before their speech motor system is ready to produce the sounds. A child who points to "more" and then "juice" on a core word board is doing the same mental work as a child saying the words out loud. That practice counts, even without sound attached.

If your child is minimally verbal or hard to understand, it's worth asking an SLP about starting AAC alongside speech practice, not instead of it.

Does how a child plays actually shape language?

Quite a bit, yes. What a child plays with, and how that play is set up, changes both how much language they hear and how much they produce themselves.

Open-ended toys pull more language than closed-ended ones. Blocks, play dough, water, sand, figurines: any of these can turn into a tower, a road, a house, or a pile that just fell down, so every new setup opens the door to new words. A shape sorter has one right answer, and then it's done.

Joint attention drives a lot of early language learning: when a child and an adult are looking at the same thing at the same moment, that's when learning really takes off. A classic Child Development study by Tomasello and Farrar found that the amount of object-focused joint attention at 18 months predicted vocabulary size at 24 months ^[8]. Getting on the floor, following where your child's eyes go, and narrating what already has their attention beats any flashcard set on the market.

Screens run into a real limit here. Even the educational ones can't offer the back-and-forth that pushes a child toward combining words, because a screen can't pause, notice what's caught a child's attention, and respond to it. A parent can.

For kids who love their screen time, the fix isn't cutting it out. Sit with them and narrate alongside what's happening: "dog run," "oh no," "ball gone." You become the responsive part of the experience the screen was never going to provide.

How many words before two-word phrases show up?

Most children start combining words on their own once they have around 50 expressive words; that's the consensus in the research ^[2].

That doesn't mean you wait until word 50 arrives before doing anything. Model phrases from day one. But if a child has fewer than 20 words, the priority is building single-word vocabulary, and two-word modeling just runs quietly in the background while that happens.

Fifty is a rough marker, not a hard wall. Some kids combine words earlier; others wait until 70 or 80 words. What matters is the direction things are moving: is vocabulary growing, does the child seem motivated to communicate, is comprehension solid?

A parent-report tool like the MacArthur-Bates Communicative Development Inventories (CDIs) gives you a reasonably reliable way to track expressive vocabulary. They're free and normed on thousands of children ^[9], and they give you a percentile and an actual word count, which beats trying to remember what your kid said last week.

If the word count keeps climbing but two-word combinations still haven't shown up by 24 months, that's the point to bring in an SLP rather than waiting another six months to see what happens.

Where early intervention fits in

Under the Individuals with Disabilities Education Act (IDEA), children under age 3 who qualify get services through Part C early intervention programs at no cost to the family ^[10]. If your child is well behind on language milestones, you can ask for a referral for evaluation. If they qualify, you'll get an Individualized Family Service Plan (IFSP) laying out speech services delivered in natural environments, which in practice usually means your own home. IDEA Part C requires that services happen "in natural environments, including the home and community settings in which children without disabilities participate, to the maximum extent appropriate" ^[10]. That means the kitchen, the backyard, or a daycare room, not a clinic. The activities in this piece are built to look like exactly what an early intervention speech-language pathologist would coach you to try. A referral doesn't have to come from a pediatrician, either. In most states, parents can self-refer straight to the Part C program. The evaluation is free, the process typically runs 45 days from referral to IFSP, and no diagnosis is required to qualify. A documented developmental delay is enough on its own. Once your child turns 3, services shift to Part B of IDEA through the school district, and the structure changes some. What stays the same is the right to an evaluation and, if your child is eligible, free services ^[10]. So don't let uncertainty about a diagnosis stop you from asking for a referral. An evaluation is how you get the answer, not something you need to arrive with already in hand.

My child repeats phrases instead of building their own

That kind of repeating, echolalia, isn't a dead end before two-word phrases show up. It can actually be a bridge to them. A lot of kids use echolalia to try out language they've absorbed. The classic example: a child says "do you want a snack?" when what they mean is that they want a snack, because that's the sentence they happened to hear in that moment. It gets the message across, even though grammatically it's borrowed rather than built from scratch. The goal isn't to stamp out the repeating. It's a slow shift from scripted, borrowed phrases toward ones the child assembles themselves, and you get there the same way you'd build any phrase: model language in the moment, stretch the script bit by bit, and give your child real reasons to want to say something. If your child echoes "want cracker" from something they heard days earlier, that's already proof they can produce a two-word phrase. From there the job is adding a third word, then a fourth, and eventually helping them swap pieces in and out on their own. Our piece on echolalia goes into more detail on how it works and what to do with it. Autistic children often lean on echolalia heavily, and the same approach still holds. Pairing spoken modeling with visual supports and AAC tends to help the shift toward flexible, self-generated phrases move faster ^[5].

Tracking progress without turning play into a quiz

You don't need a clipboard for this. The easiest approach is a running note on your phone: jot down new words and new word combinations as you hear them in real life, not the ones you asked for. Log the date, what was happening, and the phrase itself. Give it a month and you'll have something worth looking at. You're not looking for perfect sentences here. What matters is whether the number of different two-word combinations keeps growing. One new combination a week counts as real progress. A child with zero combinations in January and eight by March has made a genuine leap, even if they're still behind other kids their age. A short video clip can say more than any parent summary, since it picks up on tone, gesture, and context in a way words on a page can't. Most early intervention SLPs, and plenty of private ones too, are happy to look at a 60-second clip of play as part of an evaluation or check-in. Try to resist the urge to quiz your child. Asking "Say 'more juice'" or "What is this?" turns the moment into a test rather than a chance to learn. A child who can produce a word on command isn't necessarily in the same place as a child who says it on their own, unprompted, in the middle of play. That spontaneous use is really what you're after. If you'd rather use something structured, the MacArthur-Bates CDI mentioned earlier is free and validated ^[9]. Apps like Little Words can also help parents track language and model targets during everyday routines, with an SLP reviewing the logs to adjust things as needed. If you're not sure where to start, the quick quiz on our site is a reasonable first step.

When it's time to get a professional opinion

Some signals mean home activities alone aren't going to cut it, and waiting it out isn't the right call. A 24-month-old with fewer than 50 words and no two-word combinations should be evaluated. A 24-month-old who had words and lost them needs evaluation, full stop. And a child of any age whose speech is very hard for strangers to understand, whose vocabulary isn't growing despite a couple of months of steady effort at home, or who seems genuinely frustrated trying to communicate, should see an SLP. The AAP's Bright Futures guidelines call for developmental surveillance at every well-child visit, plus formal screening at 9, 18, and 30 months ^[11]. If your pediatrician uses a validated screener like the M-CHAT-R (for autism-specific concerns) or the Ages and Stages Questionnaire, a flagged result at any of those visits should lead straight to a referral rather than a wait-and-see approach. Cost shouldn't be what stops you: pediatric speech evaluations are covered by Medicaid in every state and by most private insurers, and Part C early intervention is free for children under 3 no matter the family's income ^[10]. An evaluation isn't a commitment to a diagnosis or a label. It's just information. An SLP will tell you whether there's something to address, what kind of issue it is, and what to do next. That's the starting point for everything after.

Frequently asked questions

What are the easiest two-word phrase activities to start with?

Sabotage play and predictable book routines have the lowest setup and the highest return. Blow bubbles and pause before giving more. Read a repetitive book until your child can anticipate the next phrase, then leave space for them to fill it in. Both run on the child's own motivation and need zero materials beyond what most families already own.

My child has about 30 words but no combinations. What should I do first?

Build vocabulary toward that 50-word threshold while modeling two-word phrases in context. Use expand-and-model routines all day: add one word to whatever the child says, no demand. At the same time, request an SLP evaluation. At 30 words without combinations, home strategies and professional input together do more than either one alone.

How long does it take for activities to show results?

Nobody has precise data on this because it swings with the child's starting point, how consistent practice is, and the underlying profile. The closest research suggests parent-implemented intervention produces measurable gains in 3 to 4 months of steady practice. Short daily sessions of 10 to 15 minutes beat one long weekly one. Expect gradual change, not a sudden breakthrough.

Should I correct my child when they say a word wrong or leave out the second word?

No. Correction tends to cut down how much a child talks, because it pairs communication with failure. Model the correct form naturally instead, without flagging the error. If your child says "milk," you say "more milk" while pouring. They got what they wanted, they heard the target phrase, and nothing felt like a mistake.

Is there a difference between a late talker and a child with a language disorder?

Yes, though the line isn't always obvious early. A late talker is usually behind in expressive language but shows normal comprehension, social engagement, and no other developmental concerns. A language disorder is broader, hitting comprehension, grammar, or social use of language. Only an SLP evaluation can tell them apart, which is why an evaluation is worth pursuing rather than waiting.

Can I use picture cards or flashcards to teach two-word phrases?

Flashcards on their own don't work well for early phrase building. They're stripped of context and lean on demand rather than real communication. A much better use of pictures: put them in a communication book or simple AAC display and point to them while speaking during real activities. That's aided language input, which has actual research behind it.

My child uses signs but not spoken words. Should I keep using signs?

Yes. Signs support spoken language rather than compete with it. Keep signing and pair each sign with the spoken word, and a symbol where you can. Multiple modes give the child several routes to the same meaning. The research is consistent that multimodal input speeds language development in late talkers and children with autism.

Are there specific toys that help children combine words?

Open-ended toys beat closed-ended ones. Play dough, building blocks, figurines, water play, and sensory bins all generate more language than shape sorters or single-function electronic toys. Action toys like wind-up cars or pop-up boxes create natural communication turns, because the child needs you to set them off again. That's a built-in reason to request.

Does watching educational shows like Sesame Street help?

Screen input on its own doesn't produce phrase combinations in late talkers. The problem is contingency: a screen can't respond to the child. What helps is watching alongside and giving live commentary. "Dog run." "Oh no, fall down." You become the responsive layer the screen can't be. Use screens with a plan, not as a stand-in for interaction.

My child has autism. Do these activities work differently for autistic kids?

The core strategies stay the same, with a few tweaks. Add visual supports and AAC alongside spoken modeling. Follow the child's interests more strictly instead of introducing new topics. Keep sessions shorter and the sensory load lower. For autistic children who use a lot of echolalia, work with an SLP on building flexible phrases alongside the scripts rather than trying to erase the echolalia. See our full guide on autism speech therapy for more.

How do I know if my child qualifies for free early intervention services?

Children under 3 with a significant developmental delay in any area, including speech and language, qualify for evaluation under Part C of IDEA at no cost to the family. Exact eligibility for services varies by state, but you don't need a formal diagnosis. A developmental delay documented by the evaluation team is enough. Parents can self-refer, so a pediatrician's referral is not required.

Is online speech therapy as effective as in-person therapy for late talkers?

The evidence for telehealth speech therapy keeps growing. A 2021 review reported in the International Journal of Language and Communication Disorders found parent-coaching models delivered by telehealth produced outcomes comparable to in-person delivery for young children with language delays. Telehealth works especially well for parent-coaching, which is already the most effective approach for this age. It's a real option, not a compromise.

What's the difference between mean length of utterance (MLU) and word count?

Word count is how many different words a child can say. MLU measures the average length of their utterances, counted in morphemes (meaning units, smaller than words). A child who says "dogs" has an MLU of 2, because "dog" and the plural marker each count. SLPs use MLU to track grammar. For home tracking, word count and number of different two-word combinations are simpler and still meaningful.